SAT in heat treatment: purpose, data flow and reporting

Standards · Last reviewed 2026-08-26
Written by Pyrolog Engineering · Reviewed by Multitek Thermal Calibration Team

Short answer: a System Accuracy Test (SAT) compares the furnace's controlling, monitoring or recording system with a traceable test system in the location and operating context defined by the governing procedure. A useful SAT record preserves both readings, corrections, identities, time and arithmetic—not only a pass/fail label.

The exact test frequency, method, permissible alternatives and acceptance limits depend on the applicable standard revision, instrumentation classification, customer flow-down and approved plant procedure. Verify those sources before planning or judging a test.

What the data workflow should capture

• furnace, zone, system instrument and channel under test;
• system sensor and test sensor/reference identities;
• calibration status and authorised correction values;
• test setpoint/condition and evidence of the required stability;
• paired readings with trustworthy timestamps;
• raw difference, applied corrections and final calculated result;
• procedure/standard revision, operator and disposition where required.

Keep the arithmetic explicit

Define the sign convention. For example, if the worksheet reports “system indication minus reference,” state that beside the result. If a sensor or instrument correction is applied, show the uncorrected value, correction sign and corrected value. This prevents a plausible number from hiding a sign error.

Worked arithmetic example

Assume an illustrative worksheet defines the reported error as corrected system value − corrected reference value. The furnace-system indication is 802.0 °C and its authorised correction, interpreted exactly as stated on the relevant record, is +0.5 °C. The test-system reading is 800.8 °C and its authorised correction is −0.3 °C.

The +2.0 °C result is not a pass/fail statement. The reviewer must use the applicable standard revision, customer requirement and approved procedure to decide disposition. If a certificate defines “correction” differently—for example, as the value to subtract—the worksheet must follow that definition rather than copy this arithmetic.

A practical sequence

1. Freeze the authority. Record the applicable revision, customer additions, local procedure and the person responsible for the decision.
2. Identify both paths. Capture the furnace, zone, instrument, channel, system sensor, test instrument and test sensor before readings begin.
3. Verify readiness. Confirm calibration status, authorised correction data, connection/position context and any required pre-test checks.
4. Reach the test condition. Record the setpoint or operating condition and the evidence required by the procedure to begin the comparison.
5. Pair readings. Use trustworthy timestamps and avoid retyping where acquisition or controlled import is possible. Make missing or invalid readings visible.
6. Calculate transparently. Preserve raw readings, correction sources, correction signs, corrected values and subtraction direction.
7. Review and retain. Apply the verified acceptance rule, capture observations/disposition, and package the worksheet with the supporting identities.

Common mistakes that create a weak SAT record

• A result with no direction: “error = 2” is ambiguous unless the report says which value was subtracted from which.
• Corrections applied invisibly: replacing the raw reading with a corrected number prevents later reconstruction and can hide a sign mistake.
• Unpaired readings: copying two values observed at different moments can turn process movement into apparent system error.
• Generic instrument names: “reference meter” is not enough when the calibration record, serial number or sensor identity must be traced later.
• Hard-coded limits: a calculator should not silently declare compliance using an obsolete edition or the wrong customer flow-down.
• Pass/fail without evidence: the disposition should link to the underlying readings, corrections, method and acceptance source.

Plant-floor scenario: the number changes after review

A technician records a system indication and a reference reading, then types only their difference into a spreadsheet. During review, the team discovers that one correction was applied with the wrong sign. Because neither raw value nor correction source is preserved, the original result cannot be reconstructed confidently. A stronger workflow stores all four numeric components, the sign convention and the calculation version; the reviewer can correct the interpretation without rewriting the measurement history.

This is where controlled automation helps most. It reduces transcription, forces required identities into view and can regenerate a report from authoritative values. It should still allow a reviewer to see the source record and explain every transformation.

What automation can and cannot do

A portable Pyrolog workflow can guide identity capture, acquire or import readings, perform transparent arithmetic, assemble a report and retain a consistent history. It cannot decide which standard or customer requirement governs without project input.

Evidence worth retaining

Keep the complete SAT record as a coherent package: authority and revision; furnace/zone/channel identity; system and test-sensor paths; calibration references; raw and corrected readings; time/context; arithmetic definition; operator observations; acceptance source; disposition; report/calculation version; and any approval required by the plant procedure. Retention and access rules should match the quality system rather than the convenience of one workstation.

Sources

• SAE International — AMS2750H Pyrometry (official revision listing; verified 2026-08-26).
• AIAG — CQI-9 Heat Treat System Assessment (official manual page; verify the governing edition before project use).

The SAT error calculator exposes the difference and user-supplied correction without declaring a standards pass/fail result.